Switch from Caspase-dependent to Caspase-independent Death during Heart Development

Differentiated cardiomyocytes are resistant to caspase-dependent cell death; however, the mechanisms involved are still uncertain. We previously reported that low Apaf1 expression partially accounts for cardiomyocyte resistance to apoptosis. Here, we extend the knowledge on the molecular basis of cardiac resistance to caspase activation by showing that the whole caspase-dependent pathway is silenced during heart development. Experimental ischemia triggers caspase activation in embryonic cardiomyocytes and proliferating fibroblasts, but not in neonatal and adult cardiomyocytes. Ischemia induces the release of the proapoptotic factors cytochrome c, truncated-AIF, and EndoG from mitochondria in postnatal cardiomyocytes in the absence of caspase activation. On the one hand, lentiviral-driven knockdown of EndoG shows that this gene is essential for ischemia-induced DNA degradation in neonatal cardiomyocytes, but not in proliferating fibroblasts; on the other hand, the AIF gene is essential for high molecular DNA cleavage in fibroblasts, but not in postmitotic cardiomyocytes, where it plays a prosurvival role during reoxygenation. These results show the switch from caspase-dependent to caspase-independent death pathways after cardiac cell differentiation, and disclose the relevance of EndoG in the caspase-independent DNA processing of differentiated cardiomyocytes.

Apoptosis, the best characterized type of programmed cell death (PCD), 6 is executed by a family of cysteinyl aspartate pro-teinases known as caspases (1). Activation of caspases, either by extrinsic or intrinsic signals, is regulated by a set of proteins, including the components of the death-inducing signaling complex (DISC), in the death receptor-mediated extrinsic pathway, and the Bcl-2 family of anti-and proapoptotic proteins, as well as the apoptosome complex including Apaf-1, in the intrinsic pathway, which involves the mitochondrion (2). The apoptotic process ends with the degradation of nuclear DNA by the caspase-dependent endonuclease CAD/DFF40 (3).
Although the order in which every step of caspase-dependent PCD takes place is well defined in proliferating and undifferentiated cells, the precise cell death pathways activated in differentiated cardiomyocytes and neurons remain more elusive (4 -6). In cardiomyocytes, the prevailing dogma that the caspase-dependent machinery is fully functional has been recently challenged (4,7). In addition, although cardiac apoptotic rates of 0.1-0.01% occurring during cardiac disease have been estimated to be pathologically relevant (8,9), the underlying mechanisms for the low incidence of apoptosis have not been addressed. Nevertheless, apoptotic cells have been detected in the heart and in cardiomyocyte cultures during ischemia (10 -13). Therefore, cardiac ischemia is a clinically relevant model for analyzing the mechanisms that promote death in differentiated cardiomyocytes.
Potential executors of caspase-independent cell death have been discovered. The mitochondrial protein apoptosis-inducing factor (AIF) was identified as an important executor of cell death when translocated to the nucleus (14). Translocation of AIF has been proposed to be both caspase-dependent (15,16) and caspase-independent (17)(18)(19), although the reason for this discrepancy is uncertain (20). Furthermore, the death-inducing role of AIF in differentiated cells is controversial. Indeed, the participation of AIF in cell death has been either confirmed (19,(21)(22)(23) or discarded (24,25). Thus, the role of AIF in promoting cell death and the mechanisms involved in promoting its release from mitochondria need to be better characterized.
In the nematode Caenorhabditis elegans, AIF is a component of the caspase-dependent mitochondrial pathway, which also involves the mitochondrial endonuclease EndoG (26). However, the role of EndoG in apoptotic DNA damage was initially suggested to be caspase-independent (27,28). Furthermore, recent data from EndoG deficient mice do not apparently support a relevant role of this protein in mammalian cell death and DNA processing (29,30). Therefore, the role of EndoG in mammalian cell death and the mechanisms involved in its release from mitochondria deserve further investigation.
While studying the mechanisms involved in cardiomyocyte cell death, we found that these cells down-regulate the expression of caspases after birth. Therefore, we checked the role of several known caspase-independent effectors of cell damage by using a lentiviral-driven gene knockdown approach. Results herein reported prove that cardiomyocytes lose the competence to dye by apoptosis after the caspase-dependent machinery is silenced during development, and show the existence of a caspase-independent, yet mitochondria-dependent, cell death pathway involving translocation and activation of EndoG, leading to the degradation of nuclear DNA, independently of AIF.
Cell Culture, Treatments, and Viability Measurements-Rat neonatal cardiomyocytes and heart fibroblasts were obtained from the heart of 3-5-day-old Sprague-Dawley rats as described elsewhere (4,31). Embryonic cardiomyocytes (E16) were obtained following essentially the same protocol, but reducing the time of digestion to 15 min, and were cultured in the same medium enriched with 1 M insulin (Sigma), 1 ng/ml cardiotrophin-1 (R&D Systems), and 10 nM mouse epidermal growth factor (Upstate). Adult cardiomyocytes (P180) were obtained and cultured following the protocol described by Ravassa et al. (32). Experimental ischemia was achieved by culturing cells in Tyrode's solution (NaCl 137 mM, KCl 2.7 mM, Na 2 HPO 4 8 mM, KH 2 PO 4 1.5 mM, CaCl 2 0.9 mM, and 0.5 mM, initial pH: 7.2) inside a hypoxic chamber (Billups-Rothenberg) in a mixture of 5% CO 2 and 95% N 2 following manufacturer's instructions to attain a 0.1% oxygen concentration (31). Quantification of cell death was performed by the trypan blue exclusion assay at the end of treatments as reported (31). Cell death in each experimental condition was measured in duplicates, and error bars represent the S.E. of three independent experiments.
Enzymatic Caspase Activity Assay-Caspase-8 (Z-IETD-AFC) and executioner (Z-DEVD-AFC) caspase activity were measured in cell extracts as previously reported (4), using the fluorogenic substrates at 50 M in 96-well plates. Fluorescence was detected with a Bio-tek FL 600 fluorometer (Izasa). Data obtained for different experimental conditions were compared within the linear phase of absorbance increase. Data are the mean of three independent experiments performed in duplicates.
Cytosolic Extracts-Cytosolic fractions were obtained as previously reported (4) at the end of the treatments to detect the release of apoptotic factors from the mitochondria.
Western Blot and Immunofluorescence Detection-Whole cell lysates were obtained by addition of Tris-2%SDS pH: 6.8 buffered solution to the cell cultures at the end of the treatments. Protein extracts were denatured by heat shock at 95°C for 3 min and quantified by the Lowry Assay (Bio-Rad). Equal amounts of protein were electrophoresed by SDS-PAGE, transferred to polyvinylidene difluoride membranes (Amersham Biosciences), and probed with specific antibodies (supplemental Table S1). For immunofluorescence detection, cells were cultured in 4-well plates (NunClon) and fixed by either incubation with paraformaldehyde 4% for 10 min or by 100% methanol for 5 min. Cells were rinsed, permeabilized, and stained as described previously (31). For apoptosis quantification, cells showing cleaved caspase-3 expression and nuclear fragmentation were counted as apoptotic. At least, 100 cells were analyzed per condition and values are mean Ϯ S.E. of three independent experiments.
DNA Integrity Assay-Cells were pelleted at the end of each treatment and frozen at Ϫ80°C. Pellets from the same experiment were processed at once. They were diluted in 40 l of sterile phosphate-buffered saline, mixed with 40 l of melted 1% low melting agarose (Sigma) in 0.5ϫ TBE (45 mM Tris pH 8.3, 45 mM boric acid, 1.0 mM EDTA). Each mixture was poured into a block caster and let to solidify. Each agarose block was submerged into 1 ml of lysis buffer (1% lauryl sarcosil, 0.5 M EDTA, 10 mM Tris, pH 8, 100 g/ml proteinase K) at 50°C during 24 h in mild agitation, and rinsed twice with 0.5ϫ TBE for 1 h at room temperature. For analysis of DNA high molecular weight degradation, the blocks were then laid into wells of a 1% agarose, 0.5ϫ TBE gel (CHEF grade, Sigma). Pulse field electrophoresis was performed in a CHEF DR-II system (Bio-Rad) set to the following protocol: run time, 14 h; switch time from 5 to 50 s; voltage gradient, 6 V/cm. Initial lysis buffer was further processed to analyze DNA low molecular weight degradation. Briefly, 1 ml of lysis buffer from DNA extraction was mixed with 1 ml of ethanol, kept at Ϫ20°C for 18 h, and centrifuged. The pellet was rinsed with 70% ethanol, centrifuged again, and the final DNA pellet was diluted in 20 l of 10 mM Tris, pH 8, 1 mM EDTA, and 10 g/ml RNase. Conventional 2% agarose-gel electrophoresis was performed. Gels were stained with SYBR Safe (Molecular Probes), visualized by UV exposure and recorded with a Kodak DC290 digital camera.
Overexpression and Detection of FLAG-tagged EndoG in Cardiomyocytes-To test the hypothesis that EndoG translocates from mitochondria to cytosol during ischemia in cardiomyocytes, the coding sequence of EndoG contained in the mouse IMAGE clone 5029633 (MRC Gene Service) was PCRamplified flanked by the EcoRI and XbaI sites and was subcloned into the pCDNA3.1 expression plasmid (Invitrogen) inframe and upstream to a FLAG tag. Cardiomyocytes were transfected by the Lipofectamine system (Invitrogen) and overexpressed EndoG-FLAG was detected by immunofluorescence and Western blot with an anti-FLAG antibody (Sigma).
Gene Knockdown by Small Hairpin RNA Interference (shRNAi) using Lentiviral Vectors-Specific 19 nucleotide sequences were chosen for each gene (AIF, EndoG) using the free RNAi design interfaces available at the Promega and Invitrogen web sites. Primer couples were designed for cloning in the pSUPER plasmid (Oligoengine) using the BglII and HindIII sites. Primers were designed against the sequences 5Ј-GGCAACATGGTGA-AACTTA-3Ј for rat AIF and 5Ј-GGAACAACCTTGAGAA-GTA-3Ј for rat EndoG. Then, an EcoRI/ClaI fragment containing the H1 promoter for the RNA polymerase III and the shRNAi sequence was cut from pSUPER and subcloned into the pLVTHM plasmid (from Dr. Trono, Geneva). For each gene, we designed a couple of primers carrying two nucleotide changes from the original sequence that were initially used to verify the specificity of the shRNAi constructs. Each specific shRNAi vector was transfected into HEK293T cells together with the plasmids psPAX2 and pMD2G (from Dr. Trono), using the polyethyleneimine transfection method (Sigma). Transfection efficiency was analyzed by detection of the enhanced green fluorescence protein (EGFP) expression, whose coding sequence is contained in the pLVTHM plasmid and driven by an IRES sequence. The 293T medium was collected after 48 h of transfection and centrifuged at 50,000 ϫ g for 3 h. The final viral pellet was diluted in sterile phosphate-buffered saline plus 2% bovine serum albumin and titrated by transduction of cardiac fibroblasts before treating cardiomyocytes. Cardiomyocytes were transduced 4 h after seeding. Efficiency of transduction was estimated to be over 80% by detection of EGFP expression at 48 h. The capability of the constructs for repressing each gene was routinely tested by Western blot for AIF and by RT-PCR from total RNA for EndoG, as described below. The maximum efficacy of the shRNAi plasmids for reducing endogenous expression of the genes was determined to reach a plateau from 4 -6 days after transduction, depending on the gene. Therefore, treatment of cardiomyocytes was started always 6 days after transduction, i.e. after seeding. All experiments carried out with transduced cardiomyocytes were repeated at least four times.
Detection of EndoG Transcript by RNA Reverse Transcription Coupled to Standard PCR-Gene expression was analyzed by reverse transcription coupled to PCR (RT-PCR) from total RNA extracts as reported for other genes (31). Conventional PCR was performed in a GeneAmp PCR System 2700 thermocycler (Applied Biosystems), using an annealing temperature of 55°C, with the primers forward: 5Ј-GACTTCCAC-GAGGACGATTC-3Ј, and reverse: 5Ј-AAGCTGCGGCTG-TACTTCTC-3Ј, producing an amplicon of 218 bp. Control for cDNA input in the amplification reaction performed with a couple of primers amplifying a fragment of the "upstream of n-Ras" (unr) gene. PCR products were migrated in 2% agarose gels and visualized by SYBR Safe staining.
Statistical Analysis-Data are expressed as mean Ϯ S.E. of three independent experiments. The significance of differences among means was evaluated using the Student's t test. A value of p Ͻ 0.05 was considered significant.

Down-regulation of the Caspase-dependent Death Pathway during Cardiac Development Is Associated with a Decrease in
Ischemia-induced Caspase Activation in Cardiomyocytes-Expression of the main regulators of caspase-dependent cell death decreased with age in the rat myocardium (Fig. 1A). Apoptotic gene repression took place also in vitro in postnatal cardiomyocytes (Fig. 1B). Experimental ischemia (ischemia, hereafter) was achieved by culturing cells inside a hypoxic chamber (31) in a Tyrode's solution without serum and glucose that becomes acidic during hypoxia (final pH 6.1-6.4). We used skin fibroblasts as positive control for the activation of caspases in this setting (31). Executioner caspase-3 activation was observed in embryonic cardiomyocytes and skin fibroblasts, but not in postnatal and adult cardiomyocytes during ischemia or stauro-  AUGUST 11, 2006 • VOLUME 281 • NUMBER 32 sporine treatment ( Fig. 1C and supplemental Figs. S1-S3). However, ischemia induced cell death to a similar extent in postnatal cardiomyocytes and skin fibroblasts, whereas cardiac fibroblasts survived to the ischemic period (Fig. 1D). On this account, we have previously reported that cardiac fibroblasts are resistant to apoptosis because of their strong expression of the antiapoptotic factor Bcl-2 (31). These data establish a correlation between the silencing of the apoptotic genes during myocardial development and the reduction in the competence of postnatal cardiomyocytes to activate apoptosis.

EndoG Executes Ischemia-induced Cardiac DNA Processing
Lack of Caspase Re-expression and Caspase Activity in Cardiomyocytes during Ischemia and Relevance of the Mitochondrial Pathway in Ischemia-induced Fibroblast Apoptosis-Expression of the initiator caspases-8 and -9, the postmitochondrial regulator Apaf-1, and the executioner caspase-3, were significantly less expressed in cultured postnatal cardiomyocytes than in cardiac and dermal fibroblasts ( Fig. 2A). Importantly, myocardial expression of these genes was not up-regulated during ischemia ( Fig. 2A). Processing of caspase-9, an initiator caspase in the mitochondrial pathway, occurred in skin fibroblasts and, later in cardiac fibroblasts, but not in cardiomyocytes ( Fig. 2A). Ini-tiator caspase-8-like activity was measured as an indication of the role of the death receptor-dependent pathway during ischemic cell death. Caspase-8 activity was not detected in ischemic cardiomyocytes, cardiac, and dermal fibroblasts before caspase-3 was engaged (Fig. 2B). Executioner caspase activity was not detected in ischemic cardiomyocytes but was evident in skin fibroblasts (Fig. 2C). These results suggest that a low expression of the caspase-dependent machinery in postnatal cardiomyocytes is maintained during ischemia, thus limiting the impact of caspase activation. Furthermore, executioner caspase processing was not substantially enhanced when ischemic adult cardiomyocyte cultures (pH 6.1-6.4) were reoxygenated and cultured in standard medium (pH 7-7.5) (Fig. 2D). However, ischemia activated the mitochondrial pathway of apoptosis in proliferating cell types such as fibroblasts (Fig. 2A).
The Pattern of DNA Degradation Induced during Ischemia in Cardiomyocytes Suggests the Participation of Caspase-independent Mechanisms-Caspase-dependent cell death induces a characteristic pattern of DNA degradation related to the activity of endonucleases that cleave DNA between nucleosomes, rendering fragments of low molecular weight, the so-called DNA laddering. Caspase-independent PCD can induce high molecular weight fragmentation of DNA, mainly in fragments of ϳ50 kb (14), and/or low molecular weight fragmentation, which appears as a smear after electrophoresis, because of the ability of the caspase-independent DNAses, such as EndoG, to cleave inside the nucleosomes (33). In aerobic conditions, serum and glucose deprivation for 20 h induced two types of DNA degradation in cardiomyocytes. A faint DNA low molecular weight laddering (Fig. 3A, lower panel) that was partially reverted by the pan-caspase inhibitor z-VAD.fmk (Fig. 3B, lower panel), and high molecular weight fragmentation (Fig.  3A, upper panel), which was not blocked by z-VAD.fmk (Fig.  3B, upper panel). During ischemia, i.e. serum and glucose deprivation in hypoxia and acidosis, cardiomyocyte DNA was fragmented rendering a low molecular weight smear in agarose gels. Upon ischemia, DNA smear was not inhibited by the caspase inhibitor z-VAD.fmk. PARP-1 has been involved in ischemia-induced neuronal cell death (34); however, ischemiainduced cardiac DNA damage was not blocked by the two PARP inhibitors DIQ and DPQ. In addition, although calpains have been shown to trigger the release of AIF in vitro (35), addition of the pan-calpain inhibitor MDL 28170 did not prevent DNA high molecular weight degradation in ischemic cardiomyocytes (Fig. 3, A and B). Furthermore, high molecular weight fragmentation of DNA, compatible with the activity of caspase-independent pathways, was rapidly activated within a few hours of ischemia (Fig. 3A, upper panel), and was not inhibited neither by z-VAD.fmk, nor by DIQ. In contrast, DNA degradation was not observed in ischemic heart fibroblasts after 24 h of treatment, in agreement with our previous results (31). Ischemic skin fibroblasts showed a canonical pattern of DNA low molecular weight degradation, which was completely abol-ished by z-VAD.fmk (Fig. 3, A and B). Therefore, the pattern of DNA degradation during cardiomyocyte ischemia and its execution in the presence of caspase inhibitors suggest the predominant role of a caspase-independent mechanism of cell death.
Ischemia Induces Cytochrome c, Truncated AIF, and EndoG Release from Cardiac Mitochondria in the Absence of Caspase Activation-We have previously reported that staurosporine does not induce caspase activation in cardiomyocytes despite cytochrome c translocation, and that overexpression of Apaf-1 partially antagonizes this resistance (4). Therefore, we analyzed the release of proapoptotic mitochondrial factors during ischemia in cardiomyocytes. Cytochrome c translocation was detected during ischemia in embryonic and postnatal cardiomyocytes (Fig. 4A, panels B, D, F, and H, and Fig. 4C); however, this event was followed by caspase-3 cleavage (activation), and by nuclear fragmentation in embryonic ( Fig. 1C and Fig. 4A, panels C and D) but not in postnatal cardiomyocytes (Fig. 4A,  panels G and H).
AIF and EndoG are mitochondrial proteins that have been reported to act as executioners of PCD after their translocation to the cytosol (14,27). AIF was released from mitochondria during cardiac ischemia (Fig. 4, B and C), but cytosolic AIF was smaller in size than mitochondrial AIF, as detected with an antibody raised against the C terminus of the protein (Fig. 4C). This is in agreement with the recently proposed apoptosis-induced cleavage of AIF at the N terminus, which allows its detachment from the inner mitochondrial membrane and subsequent translocation (36). This cleavage has been proposed to depend in vitro on calpain and/or cathepsin activity (35,36). However, pretreatment of cardiomyocytes with the pan-cal-  pain inhibitor MDL 28170 did not prevent AIF cleavage and its release during ischemia (Fig. 4C).
Ischemia-induced EndoG translocation was investigated by immunofluorescence and by Western blot with three commercial antibodies (supplemental Table S1). However, our results suggested that these antibodies are not specific when used on rat samples. We therefore constructed a plasmid for the overexpression of mouse EndoG with a C-terminal FLAG tag. EndoG-FLAG tracking demonstrated that EndoG had a punctuated pattern that excluded the nucleus under basal conditions, which is consistent with a mitochondrial localization (Fig. 5A, panels A and B). However, EndoG pattern changed during ischemia in accordance with its translocation to the cytosol and nucleus (Fig. 5A, panels C and D). We verified by Western blot that overexpressed EndoG had the expected size of 32 kDa, and no cleavage bands were observed in ischemic cardiomyocyte extracts (Fig. 5B), suggesting that EndoG translocation was not dependent on its cleavage, contrary to AIF (Fig. 5B). Taken together, these results demonstrate that mitochondrial proteins cytochrome c, truncated AIF and EndoG are translocated to the cytosol in ischemic postnatal cardiomyocytes, yet caspases and calpains are not essential for their release.
EndoG Is an Important Executor of DNA Processing during Ischemia in Cardiomyocytes-The pattern of DNA cleavage in ischemic postnatal cardiomyocyte indicated the participation of caspase-independent mechanisms. AIF and EndoG have been suggested to execute cell death in the absence of caspase activation (14,27). This prompted us to test if AIF and EndoG, which were released from mitochondria during ischemia, were responsible for cardiac cell death in this model. Cardiomyocytes were transduced with lentiviral vectors carrying constructs for small hairpin-based RNA interference (shRNAi) of either protein. Six days after transduction, cardiomyocytes were exposed to ischemia (see "Experimental Procedures"). The effectiveness of AIF shRNAi vector was evidenced by the decrease in AIF protein levels measured in cardiomyocyte total lysates (Fig. 6A). A time course analysis of AIF expression in cardiomyocytes transduced with the AIF-specific shRNAi plasmid showed that maximal reduction of AIF expression was achieved at day six after infection (Fig. 6A, upper panel). Therefore, all experiments were carried out at day six of transduction. AIF expression was reduced up to 80% in cardiomyocytes (Fig.  6A, lower panel). Reduction in EndoG expression was checked by RT-PCR from total RNA extracts to confirm the efficacy of the EndoG-specific shRNAi construct (Fig. 6B). Reduction in EndoG expression, but not in AIF levels, correlated with an important reduction of DNA damage (Fig. 6C). A faint ladder pattern was observed after electrophoresis of DNA from ischemic cardiomyocytes in which EndoG expression was reduced. This faint ladder was inhibited by z-VAD.fmk addition, suggesting a very low contribution of caspases (Fig. 6C). Interestingly, ischemia-induced DNA high molecular weight fragmentation was not blocked in cells with reduced EndoG expression, suggesting the presence of another yet unidentified AIF-independent, EndoG-independent nuclease working in the absence of caspase-activation. In contrast, AIF was essential for the ischemia-induced high molecular weight processing of DNA in skin fibroblasts (Fig. 6D), whereas the caspase-dependent DNase activity, but not EndoG, was involved in the low molecular weight DNA cleavage in these cells (Fig. 6D).
To ascertain whether the reduction in EndoG expression favored cardiomyocyte cell survival during ischemia, analysis of cell survival in scrambled-transduced, AIF-deficient and EndoG-deficient cardiomyocytes was performed. Neither AIF nor EndoG knockdown prevented ischemia-induced cardiac cell death, although reduction in AIF expression substantially blocked cardiac cell recovery during reoxygenation (Fig. 7). Altogether, these results reveal a main role of caspase-independent mechanisms in cardiomyocyte ischemic cell death, and identify EndoG as an essential executor of ischemia-induced DNA processing in these cells. EndoG-FLAG was transfected in cardiomyocyte cultures as described under "Experimental Procedures." A, expression of exogenous EndoG-FLAG was examined by immunofluorescence with an anti-FLAG antibody in control cardiomyocyte cultures 3 days after transfection (panels A and B), and in cardiomyocytes exposed to 16 h of ischemia the 3rd day (panels C and D). B, EndoG-FLAG was detected in SDS-PAGE-electrophoresed total protein extracts and cytosolic protein extracts in control and ischemic cardiomyocytes. AIF and cytochrome c were detected to check the effectiveness of the ischemic treatment in releasing endogenous mitochondrial proteins. COXIV is a mitochondrial marker and LDH is a marker for the enrichment in cytosolic proteins. The images are representative of three independent experiments.

DISCUSSION
The results presented here show that the silencing of the caspase-dependent machinery during cardiac differentiation (both in vivo and in vitro) accounts for the lack of caspasedependent cell death during experimental ischemia. Our data also prove that the main pathway for DNA processing in cardiac cells involves the caspase-independent release of the mitochondrial endonuclease EndoG, which helps other caspase-independent, AIF-independent, DNAses in the cleavage of cardiac DNA. Furthermore, our results suggest the presence of a previously not described caspase-independent nuclease activity that induces high molecular weight fragmentation of DNA in ischemic cardiomyocytes. Finally, the prosurvival role of AIF in cardiomyocytes during reoxygenation is described.
During an ischemic insult, postnatal cardiomyocytes do not activate caspases as readily as other cell types (9,37). However, the underlying molecular mechanisms are not well understood because cardiac cells have been assumed to express a functional caspase-dependent pathway. Several molecules have been described to protect cardiomyocyte mitochondria by blocking cytochrome c release, and have been proposed to be apoptotic blockers in cardiomyocytes. Heat shock proteins have been suggested to be endogenous protectors against ischemic damage in cardiomyocytes (38,39) by protecting mitochondrial function (40). Experimental overexpression of Bcl-2 or Bcl-X L , which block cytochrome c translocation, have been reported to be cardioprotective during ischemia/reoxygenation (41)(42)(43). Accordingly, the hearts of mice deficient for Bax, which induces cytochrome c release, are partially protected during ischemia (44). Apoptosis Repressor with Caspase Recruitment Domain (ARC) is highly expressed in the heart (45), and protects this organ against ischemic-induced cell death (46). ARC blocks cytochrome c release by interfering with Bax (47,48). From the above-mentioned observations, it has been assumed that activation of the mitochondrial apoptotic pathway leading to executioner caspase activation is relevant in the cardiac ischemic damage. However, caspase activity was not measured (46), or it was only analyzed in the heart-derived cell line H9c2 (45,47,48). Importantly, heart-derived cell lines H9c2 and HL-1 (49) are not terminally differentiated cells and, contrary to cardiomyocytes, express all the apoptotic regulators. 7 Furthermore, it is unknown why in conditions where caspasedependent programmed cell death should be engaged (i.e. after cytochrome c release (50)), there was only low caspase activation in cardiomyocytes (51). What blocks cardiomyocyte apoptosis after cytochrome c release? We have previously described that cardiomyocytes lack significant Apaf-1 expression (4), and this could hamper the activation of caspase-9 after 7 N. Bahi, J. Zhang, and D. Sanchis, unpublished data. , and expression of AIF in 10 and 5 g of protein from total extracts of normal cardiomyocytes (No RNAi), scrambled-transduced (scr), AIF-deficient (AIF) and EndoG-deficient (EndoG) cardiomyocytes, 6 days after infection. Hsp60 was used as loading control. B, EndoG transcript levels in scrambled-transduced cardiomyocytes (scr) or EndoG-deficient cardiomyocytes (EndoG). RT-PCR was performed from 1 and 4 g of total RNA and equal volumes of the PCR product extracted at PCR cycles 27, 30, and 35 were loaded and electrophoresed in a SYBR Safe-stained 3% agarose gel. PCR reaction from 1 g of total RNA was conducted in parallel for amplification of the unr transcript, used as loading control. C, DNA high and low molecular weight fragmentation (HMWF and LMWF) induced by exposure to 18 h of ischemia with (ϩ) or without 100 M z-VAD.fmk (zV), in scrambled-transduced (scr), AIF-deficient (AIF), and EndoG-deficient (EndoG) cardiomyocytes. Arrow indicates band at 50 kb. D, DNA high and low molecular weight fragmentation (HMWF and LMWF) induced by exposure to 18 h of ischemia in scrambledtransduced (scr), AIF-deficient (AIF ), EndoG-deficient (EndoG), and AIF/EndoG double-knockdown skin fibroblasts. The images are representative of three independent experiments. cytochrome c release. A recent report proposes that the low expression of Apaf-1 helps inhibitor of apoptosis proteins (IAPs) to better counteract executioner caspase activation in cardiomyocytes, which were suggested to express caspases at a similar level than fibroblasts (52). Contrary to this work, the data reported here demonstrate a global down-regulation of the whole caspase-dependent machinery in the heart during development. Reduction in caspase expression is evident when comparing pure isolated embryonic and adult cardiomyocytes, 7 suggesting that most of the caspase expression in neonatal cardiomyocyte cultures comes from dividing cardiac fibroblasts, which usually contaminate cardiomyocyte cultures. In addition, our results discard an inhibition of caspase activation by ischemia-induced acidosis, because fibroblasts and embryonic cardiomyocytes do activate caspases in this setting, even at pH 6.1. Therefore, we propose that differentiation-related repression of the whole caspase-dependent machinery accounts for postnatal cardiac protection against caspase activation. We hypothesize that down-regulation of the apoptotic machinery during differentiation can be a mechanism for reducing caspase-induced cell damage by accidental cytochrome c translocation due to the high number of mitochondria and the constant change of cardiomyocyte volume during the contraction/ relaxation cycle.
Our results demonstrate that cytochrome c translocation during cardiac ischemia is not linked to the caspase-dependent death pathway, contrary to the dogma that has been established in recent years. Nevertheless, mitochondrial damage is involved in ischemia-induced cardiomyocyte cell death and hampers cardiac recovery during reoxygenation. In this respect, truncated-AIF translocation from mitochondria was detected in ischemic cardiomyocytes. Of note, pharmacological calpain inhibition did not prevent ischemia-induced AIF cleavage and release. This fact discards an essential role of calpains in AIF translocation and release in ischemic cardiomyocytes. In our experimental model, AIF release did not account for ischemia-induced cardiac DNA damage. Furthermore, we show that AIF depletion hampers cardiac survival during reoxygenation. Importantly, the ischemic lesion is bigger in the AIF-defective Harlequin mouse (25). In addition, we have recently shown that AIF is not related to DNA damage in staurosporine-treated neuroblastoma cells (53). Therefore, our data suggest that AIF role in DNA cleavage depends on the cell type and does not support a relevant role for AIF in cardiac DNA processing.
The present study also reveals that EndoG is the main contributor for the DNA low molecular weight degradation in ischemic cardiomyocytes. Interestingly, two recent works reported no obvious effects on cell death and DNA degradation in EndoG-deficient mice (29,30). However, DNA integrity analysis was studied in splenocytes, which otherwise activate caspase-dependent apoptosis (29). It must be taken into account that Li et al. showed the EndoG role in mammalian apoptotic DNA degradation in fibroblasts deficient for caspaseactivated DNase (27). Furthermore, here we show that EndoG silencing in primary skin fibroblasts, which express caspases, does not induce a reduction in DNA cleavage. Taken together, these results suggest that EndoG cleaves DNA in systems where caspases are absent, irrespective of the death stimulus, and could be the final step of DNA processing in postmitotic cells.
Finally, our results show that high molecular weight fragmentation of DNA is caspase-independent in dividing fibroblasts and postmitotic cardiomyocytes and that AIF is involved in this step of DNA degradation only in proliferating skin fibroblasts but not in cardiomyocytes. Indeed, a functional genomic approach in C. elegans has highlighted the involvement of many genes in the final apoptotic DNA degradation process, which could be controlling the final cell fate (54). Therefore, we are currently investigating the identity of the nuclease activities involved in the ischemia-induced DNA high molecular weight fragmentation.
In summary, the involvement of caspases in cardiac cell death has been based on the assumption that the caspase-dependent machinery is expressed in differentiated cardiomyocytes and has been inferred mainly from the experimental evidence of cytochrome c translocation and the cleavage of DNA rendering 3Ј-OH ends, which are detected by the TUNEL assay, a characteristic feature of the caspase-dependent DNase CAD/ DFF40, but that can be also produced by EndoG (55). In contrast, here we show that cardiomyocytes reduce the expression of the caspase-dependent cell death machinery during differentiation and that these proteins are not up-regulated during ischemia. Our results show that, after differentiation, caspaseindependent mechanisms substitute for caspases in the ischemic cell damage of cardiomyocytes. Furthermore, our data disclose the essential role of EndoG, which works in concert with a yet unidentified nuclease activity, as an important executor of caspase-independent DNA cleavage in differentiated cardiomyocytes.